PULSE DURING MENTAL AND PHYSICAL WORK. 199 



Our attempt to measure the effect of the association experiment on 

 the pulse-rate rests on a similar theoretical basis. If the pulse data 

 are arranged according to their experimental incidence it may be 

 assumed in this case, as in the case of the pulse-rate, that in a sufficient 

 number of instances the non-experimental rhythms and the accidental 

 variations will tend to balance and leave only the significant experi- 

 mental change. In other words, in the case of the pulse, as in our 

 general statistical procedure, we postulate that chance variations can 

 not obscure any systematic change in the measurement of a process 

 if the number of cases be sufficiently large. In the measurement of 

 the pulse-cj^cles during association, the non-experimental rhythms are 

 treated as chance variations. Significant variations would be such as 

 correlate with the reaction process. A comparison of the average of 

 all the pulse-cycles which occur just after the moment of stimulation and 

 the average of all the pulse-cycles which occur just before that moment 

 should give the pulse correlate of the effect of stimulation. While this 

 theory of pulse elaboration is believed to be sound, it may well be 

 questioned if 50 cases are sufficient for the non-experimental rhythms 

 to be eliminated. Our only answer to that objection is that we have no 

 means of knowing. Fifty cases is, however, the best available unit in 

 our experiments, and it is not seriously different from a widely used 

 physiological standard, viz, of pulse-rate per minute. 



The experimental pulse-changes in association tests for Subject VII, 

 elaborated according to the foregoing theory, are summarized in table 

 35. The first column shows the kind of experiment and the number of 

 the word series. The average values of the pulse cycles are entered in 

 the appropriate columns under pre-stimutation pulse-cycles, stimulation 

 to reaction, and post-stimulation pulse-cycles, corresponding to the 

 arrangement of table 34. Thus the right-hand column under pre-stim- 

 ulation pulse-cycles shows the average duration of the pulse-cycles just 

 before the stimulus words were given for each period of the four experi- 

 mental days. An average pulse-rate for each experimental period is 

 shown in the extreme right-hand column. 



An examination of table 35 shows that in each normal experimental 

 session the average length of the pulse-cycles increases from period to 

 period. What is true of the average is true also at each stage of the 

 experimental cycle, say at the first post-reaction pulse. The same 

 phenomenon appears also on the second normal day. It is less con- 

 spicuous after dose A of alcohol, and is often reversed after dose B. 

 That is to say, in Subject VII alcohol tends to prevent the retardation 

 of the pulse which occurs in a " normal" 3-hour experimental session. 

 A second clear indication of table 35 is that there is a conspicuous 

 difference in the course of pulse-changes from pre-stimulation to post- 

 reaction between the normals of the day, 1, 6, 11, 16, and subsequent 

 periods of the same day. 



